During the development of a pharmaceutical formulation, a powerful tool is needed to extract the key points from the complicated process parameters and material attributes. Artificial neural networks (ANNs), a promising and more flexible modeling technique, can address real intricate questions in a high parallelism and distributed pattern in the manner of biological neural networks. The data mined and analyzing based on ANNs have the ability to replace hundreds of trial and error experiments. ANNs have been used for data analysis by pharmaceutics researchers since the 1990s and it has now become a research method in pharmaceutical science. This review focuses on the latest application progress of ANNs in the prediction, characterization and optimization of pharmaceutical formulation to provide a reference for the further interdisciplinary study of pharmaceutics and ANNs.

Purpose: Proper taste-masking formulation design is a critical issue for instant-dissolving tablets (IDTs). The purpose of this study is to use the electronic tongue to design the additives of the 3D printed IDTs to improve palatability. Methods: A binder jet 3D printer was used to prepare IDTs of levetiracetam. A texture analyzer and dissolution apparatus were used to predict the oral dispersion time and in vitro drug release of IDTs, respectively. The palatability of different formulations was investigated using the ASTREE electronic tongue in combination with the design of experiment and a model for masking bitter taste. Human gustatory sensation tests were conducted to further evaluate the credibility of the results. Results: The 3D printed tablets exhibited rapid dispersion ( 90%). The electronic tongue had an excellent ability of taste discrimination, and levetiracetam had a good linear sensing performance based on a partial least square regression analysis. The principal component analysis was used to analyze the signal intensities of different formulations and showed that 2% sucralose and 0.5% spearmint flavoring masked the bitterness well and resembled the taste of corresponding placebo. The results of human gustatory sensation test were consistent with the trend of the electronic tongue evaluation. Conclusions: Owing to its objectivity and reproducibility, this technique is suitable for the design and evaluation of palatability in 3D printed IDT development.

Chitosan is a natural polysaccharide, mainly derived from the shell of marine organisms. At present, chitosan has been widely used in the field of biomedicine due to its special characteristics of low toxicity, biocompatibility, biodegradation and low immunogenicity. Chitosan nanoparticles can be easily prepared. Chitosan nanoparticles with positive charge can enhance the adhesion of antigens in nasal mucosa and promote its absorption, which is expected to be used for intranasal vaccine delivery. In this study, we prepared chitosan nanoparticles by a gelation method, and modified the chitosan nanoparticles with mannose by hybridization. Bovine serum albumin (BSA) was used as the model antigen for development of an intranasal vaccine. The preparation technology of the chitosan nanoparticle-based intranasal vaccine delivery system was optimized by design of experiment (DoE). The DoE results showed that mannose-modified chitosan nanoparticles (Man-BSA-CS-NPs) had high modification tolerance and the mean particle size and the surface charge with optimized Man-BSA-CS-NPs were 156 nm and +33.5 mV. FTIR and DSC results confirmed the presence of Man in Man-BSA-CS-NPs. The BSA released from Man-BSA-CS-NPs had no irreversible aggregation or degradation. In addition, the analysis of fluorescence spectroscopy of BSA confirmed an appropriate binding constant between CS and BSA in this study, which could improve the stability of BSA. The cell study in vitro demonstrated the low toxicity and biocompatibility of Man-BSA-CS-NPs. Confocal results showed that the Man-modified BSA-FITC-CS-NPs promote the endocytosis and internalization of BSA-FITC in DC2.4 cells. In vivo studies of mice, Man-BSA-CS-NPs intranasally immunized showed a significantly improvement of BSA-specific serum IgG response and the highest level of BSA-specific IgA expression in nasal lavage fluid. Overall, our study provides a promising method to modify BSA-loaded CS-NPs with mannose, which is worthy of further study.

Azithromycin (AZI) is one of the most commonly used macrolide antibiotics in children, but has the disadvantages of a heavy bitter taste and poor solubility. In order to solve these problems, hot-melt extrusion (HME) was used to prepare azithromycin amorphous solid dispersion. Preliminary selection of a polymer for HME was conducted by calculating Hansen solubility parameter to predict the miscibility of the drug and polymer. Eudragit^® RL PO was chosen as the polymer due to its combination of taste-masking effect and dissolution. Moreover, the solubility was improved with this polymer. Design of experiments (DoE) was used to optimize the formulation and process, with screw speed, extrusion temperature, and drug percentage as independent variables, and content, dissolution, and extrudates diameter as dependent variables. The optimal extrusion parameters were obtained as follows: temperature—150 °C; screw speed—75 rpm; and drug percentage—25%. Differential scanning calorimetry (DSC) and Powder X-ray Diffraction (PXRD) studies of the powdered solid dispersions showed that the crystalline AZI transformed into the amorphous form. Fourier transform infrared spectroscopy (FTIR) results indicated that the formation of a hydrogen bond between AZI and the polymer led to the stabilization of AZI in its amorphous form. In conclusion, this work illustrated the importance of HME for the preparation of amorphous solid dispersion of AZI, which can solve the problems of bitterness and low solubility. It is also of great significance for the development of compliant pediatric AZI preparation.